Abstract

I know, AI. Another person talking about AI. Everyone is mad about AI, but should they be? Surely using AI, the infamous ‘black box’ (not like in an aeroplane, because those famously tell you everything that was going on quite clearly, but more in a ‘I can’t see what’s going on in here!’ sense), undermines the scientific pillars of reproducibility and explainability? For this seminar, all AI apathy is to be deposited in the box by the door.

I plan to talk about a few things: Firstly, the implications of AI in conservation, based on the findings of an AI horizon scan I conducted; Secondly, an insight into an exciting collaborative project with the Department of Computer Science on how we are using large language models in the Conservation Evidence group to find evidence of conservation actions; and lastly, how the publication of ever larger numbers of convincing-fictitious academic papers represents an existential crisis for evidence synthesis, for which it is currently woefully unprepared. Do we have an answer?